Proton pump inhibitors as risk factor for metabolic syndrome and hepatic steatosis in coeliac

disease patients on gluten-free diet

Abstract
Background Recent research has shown that patients with coeliac disease (CD) are at risk of
developing metabolic syndrome (MS) and hepatic steatosis (HS) after commencing a gluten-
free diet (GFD). This study aimed to evaluate the predictive factors for MS and HS in CD
after 1 year of GFD.

Methods
All consecutive newly diagnosed CD patients were enrolled. We prospectively collected data
about BMI; waist circumference; blood pressure; cholesterol; triglycerides, glucose and
insulin blood levels; insulin resistance (through the homeostatic model assessment HOMAIR)
and treatment with proton pump inhibitors (PPI). Diagnosis of MS was made in accordance
with current guidelines and HS was diagnosed by ultrasonography. The prevalence of MS
and HS was re-assessed after 1 year of GFD. A logistic regression analysis was performed to
identify risk factors for MS and HS occurrence after 1 year of GFD.

Results Of 301 patients with newly diagnosed CD, 4.3% met criteria for diagnosis of MS and
25.9% presented with HS at the time of CD diagnosis; 99 subjects (32.8%) had long-term
exposure to PPI during the study period. After 1 year, 72 (23.9%) patients had developed MS
(4.3 vs 23.9%; p\0.001, OR 6.9) and 112 (37.2%) had developed HS (25.9 vs 37.2%; p\0.01,
OR 1.69). At multivariate analysis, high BMI at diagnosis (OR 10.8; p\0.001) and PPI
exposure (OR 22.9; p\0.001) were the only factors associated with the occurrence of MS;
HOMA-IR (OR 9.7; p\0.001) and PPI exposure (OR 9.2; p\0.001) were the only factors
associated with the occurrence of HS. Conclusions PPI exposure adds further risk of
occurrence of MS and HS for patients with CD on GFD. The use of PPI in patients with CD
on GFD should be limited to strict indications.

Keywords Coeliac disease _ Proton pump inhibitors _Metabolic syndrome _ Hepatic

steatosis.

Introduction
Coeliac disease (CD) is the most common autoimmune enteropathy in the Western countries,
triggered by the exposure to gluten in genetically susceptible subjects [1]. Recent research
has shown that weight changes are common in patients suffering from CD after commencing
a gluten-free diet (GFD) [2, 3] and that these changes can contribute to the development of
hypertension, dyslipidaemia, diabetes mellitus, metabolic syndrome (MS) and hepatic
steatosis (HS) [4]. In this population, hepatic steatosis and non-alcoholic fatty liver disease
(NAFLD) are believed to be the expression of MS in the liver [5].

Previous research carried out by our team [6] reported a prevalence of MS of about 29% in
CD patients who had followed a GFD regimen for 1 year and that 65% of patients with MS
also presented with HS.

The increase in body weight could be due to the improvement in intestinal absorption (caused
by the exclusion of gluten from the diet) in subjects who are in a compensative hyperphagic
state. The increased total caloric intake and macronutrient composition of the diet may also
contribute to overweight and obesity in patients with CD [6].
Several recent studies have advocated the role of intestinal permeability, gut microbiota and
medication in the pathogenesis of metabolic syndrome, hepatic steatosis and NAFLD [7–11].
In particular, NAFLD in humans may be associated with increased gut permeability, possibly
caused by loosening of the intercellular junctions in the intestine, which may play an
important role in the pathogenesis of hepatic fat accumulation [7]. Additionally, the liver is
constantly exposed to products of the gut microbiota that can activate hepatic toll-like
receptor 4 (TLR4), which has been implicated in the development of liver inflammation and
fibrosis, and even hepatocellular carcinoma [8–11].

Recent studies have suggested that long-term treatment with proton pump inhibitors (PPI) is
associated with excessive weight gain [12, 13]. A suboptimal weight loss after gastric bypass
bariatric surgery was also found in two separate cohorts of PPI users [14]. Acid homeostasis
in the gut may thus play a critical role in metabolic regulation.

At present, data on predictive factors other than GFD itself of metabolic syndrome and
hepatic steatosis in CD are still scarce. We therefore decided to carry out a study to evaluate
the predictive factors for MS in patients suffering from CD after 1 year of GFD. The
secondary aim of the study was to assess which factors could predict the development of HS
in the same population.

Materials and methods

Study population and study design
Starting from a previous study on MS and HS in CD [6], including 98 patients during the
years 2012–2013, we decided to extend, amplify and better characterize the main risk factors
for MS and HS in CD. Thus, between September 2013 and March 2017 we carried out an
observational longitudinal prospective study enrolling all consecutive adult CD patients who
were diagnosed at our gastroenterology unit (tertiary centre for CD and food intolerance) at
Federico II University in Naples, Italy and retained in the study those who completed a 1-year
followup period. CD diagnosis was made in the presence of Marsh 2 histology or higher
associated with both anti-tissue transglutaminase antibodies (a-tTG) IgA greater than 7 U/
mL and positive anti-endomysial antibodies (EMA) [1].

Furthermore, a population of consecutive patients suffering from gastro-oesophageal reflux

disease (GERD), diagnosed in accordance with current guidelines [15], who referred to our
unit, was retrospectively enrolled as a ‘‘control group’’.

All patients with CD (and those with GERD) underwent the following measurements and
investigations at the time of diagnosis and after 1 year of follow-up: weight (kg), height (cm),
waist circumference (cm), blood pressure (mmHg), blood cholesterol (mg/dL), blood
triglycerides (mg/dL), blood glucose (mg/dL), blood insulin (mIU/mL), blood levels of
aspartate and alanine aminotransferase (AST, U/L and ALT, U/L) and abdominal/hepatic
ultrasonography (US). The degree of insulin resistance (IR) as per the homeostatic model
assessment index (HOMA-IR) was calculated with the following formula: HOMAIR =
[fasting blood glucose (mg/dL)] 9 [fasting blood insulin (mIU/L)/405]. The intake of and
particularly the long-term (over 6 months) exposure to PPIs after CD diagnosis were also
recorded. Blood pressure measurements complied with European and American guidelines
[16, 17].
Adherence to GFD was assessed by measuring a-tTG levels [18] and by administrating a
specific questionnaire [19] at 6 and 12 months after starting GFD. The dietary advice was
provided by a departmental dietitian with a patient-tailored approach (even including heart
health information) on the basis of daily caloric needs.

MS and HS diagnosis

In line with the MS diagnostic criteria for European and American countries [20, 21], MS
was diagnosed when waist circumference was at least 94 cm (men) or at least 80 cm (women)
and two of the following risk factors were present: blood pressure at least 130/85 or patient
on antihypertensive treatment; HDL cholesterol less than 40 mg/ dL (male patients) or less
than 50 (female patients) or patient on cholesterol-lowering treatment; blood glucose at least
100 mg/dL or patient previously diagnosed with type 2 diabetes; blood triglycerides at least
150 mg/dL or patient on triglyceride-lowering treatment. In accordance with the WHO
STEPwise approach to Surveillance (STEPS), waist circumference was measured at the
approximate midpoint between the lower margin of the last palpable rib and the highest point
of the iliac crest [22]. All serological investigations were performed in the morning after
fasting for at least 12 h, at the laboratory facilities of our hospital. The body mass index
(BMI) (kg/m2) was calculated for all patients according to the WHO criteria and duly
recorded [23].

Hepatic steatosis was diagnosed non-invasively by abdominal ultrasonography (US). The

ultrasonographic procedure was performed at diagnosis and after 1 year of GFD by using a
3.5 to 5 MHz multi-frequency probe (Logiq 7 Pro, GE Medical System, Milan, Italy).
Ultrasound examination was performed by an operator (AR) with vast expertise in
abdominal/hepatic ultrasonography. HS was diagnosed in accordance with well established
criteria [24, 25]: grade 1 (liver attenuation slightly inferior to that of spleen); grade 2 (more
pronounced difference between liver and spleen and intrahepatic vessels not seen or slightly
higher attenuation than liver); grade 3 (markedly reduced liver attenuation with a sharp
contrast between liver and intrahepatic vessels).

Statistical analysis

Data were analysed using the Statistical Package for Social Sciences (SPSS software v.15.0)
for Windows. The descriptive statistics used included calculation of mean values and
standard deviation (SD) of the continuous variables, and the percentages and proportions of
the categorical variables. Data were compared by using Student’s t test, Chi-square test,
Fisher’s exact test and ANOVA, as appropriate. The odds ratio (OR) for quantifying the
statistical difference between the dichotomous variables was also calculated. Furthermore, a
binary logistic regression was used to examine the relationship between metabolic syndrome
and hepatic steatosis as dependent variables and their possible predictors as independent
variables. The model was performed by using the stepwise backward method (Wald). The
continuous variables (BMI, waist circumference, HOMAIR, hypertension, glycaemia,
transaminasemia, cholesterolaemia, tryglyceridemia) were dichotomised in terms of being
normal or abnormal (yes/no) in accordance with the current definitions. The coefficients
obtained from the logistic regression analysis were also expressed in terms of odds of event
occurrence.

A p value of less than 0.05 was considered statistically significant.

Results
We enrolled a total of 306 consecutive adult patients with newly diagnosed CD (98 included
in the previous study). Of these, five patients were excluded from the study: three because of
type 1 diabetes mellitus and two because of ongoing pregnancy. We therefore included 301
patients with CD and carried out our analysis of their characteristics at the time of diagnosis
and after 1 year of GFD.

Baseline features of the study population at the time of CD diagnosis are summarized in
Table 1. Notably, 13 patients (4.3%) met the diagnostic criteria for MS, whilst 78 patients
(25.9%) presented with HS on ultrasonographic imaging; insulin resistance was present in 36
subjects (11.9%) and 4 patients (1.3%) suffered from hypertension.
One hundred and thirty-six patients (45.1%) were being treated with PPIs at the time of CD
diagnosis, but only 99 (32.8%) continued PPI treatment for at least 6 months during the study
period. The indication for prescription of PPI and the type of PPI taken are reported in Table
2.

After 1 year of follow-up, 72 patients (23.9%) had developed MS (4.3 vs 23.9%; p\0.001, OR
6.9; 95% CI 3.73–12.77) (Fig. 1); 112 (37.2%) had developed HS (25.9 vs 37.2%; p\0.01, OR
1.69; 95% CI 1.19–2.39); 60 (19.9%) had hypertension (19.9 vs 1.3%; p\0.001; OR 18.4;
95% CI 6.62–51.58); and 83 (27.5%) presented with insulin resistance (27.5 vs 11.9%;
p\0.001, OR 2.8; 95% CI 1.82–4.30).

A logistic regression was carried out to ascertain the effect of specific variables (age at the
time of diagnosis, gender, BMI at baseline, waist circumference at baseline, HOMA-IR,
hypertension, hyperglycaemia, hypertransaminasemia, hypercholesterolaemia,
hypertriglyceridemia and PPI treatment) on the likelihood of developing MS and HS.

In the multivariate logistic regression analysis and after adjustment for all other variables in
the model, high BMI at diagnosis (OR 10.8; 95% CI 4.2–22.1; p\0.001) and PPI exposure
(OR 22.9; 95% CI 9.2–46.5; p\0.001) were the only factors associated with MS; HOMA-IR
(OR 9.7; 95% CI 3.6–21.3; p\0.001) and PPI exposure (OR 9.2; 95% CI 4.2–21.8; p\0.001)
were the only factors associated with the development of HS. Tables 3 and 4 summarize the
results for MS and HS, respectively.

Finally, when we compared the metabolic features between users and non-users of PPI after
12 months of GFD, we found that PPI users had higher BMI, greater waist circumference,
higher glucose, cholesterol and triglycerides blood levels, higher insulin and HOMA index
value (p\0.001). Moreover, PPI users had higher risk of insulin resistance than non-users of
PPI (64 vs 9.4%; p\0.001; OR 17; 95% CI 9.1–31.9); greater rate of hypertension (38 vs 6%;
p\0.001; OR 9.8; 95% CI 4.8–20); and higher prevalence of HS (74 vs 18.8%; p\0.001; OR
12; 95% CI 7.1–22.6) and MS (60 vs 6.4%; p\0.001; OR 21; 95% CI 10.7–42.7). Table 5
clarifies these findings. Interestingly, no differences were seen between PPI users and non-
users at baseline (Table 1).

Moreover, a subanalysis between non-users of PPI, PPI users for a period of less than 6
months (PPI discontinuation) and PPI users for a period of greater than 6 months showed that
only the long-term exposure to PPIs ([6 months) was a risk factor for MS and HS (Table 6).
No statistically significant difference was seen in terms of type of PPI (omeprazole,
pantoprazole, rabeprazole, lansoprazole, esomeprazole) used and PPI dosage and in terms of
compliance to diet, as assessed by a-tTG levels and specific questionnaire between PPI users
and nonusers; furthermore, there was no significant difference in the micro and macronutrient
intake between PPIs users and non-users.

Regarding the control group, we retrospectively enrolled 50 consecutive patients with GERD
with known metabolic features both at baseline and after 1 year. Among these, 22 subjects
continued PPI for a period of greater than 6 months, while the remaining 28 patients
discontinued this class of drugs. Interestingly, no differences were seen between PPI users
and non-users at baseline. In particular, the prevalence of MS in the control group at baseline
was 8% (four patients). After 1 year of follow up, 11 subjects (22%) developed MS (p = 0.04;
OR 3.2; 95% CI 1.12–5.21) (Fig. 1). Specifically, 8 PPI users patients (36%) vs 3 non-users
(10.7%) had developed MS (p = 0.03, OR 4.7; 95% CI 1.08–20.9); 14 PPI users (63%) vs 9
non-users (32%) had developed HS (p = 0.02, OR 3.69; 95% CI 1.13–11.9); 7 PPI users
(31%) vs 2 non-users (7%) had hypertension (p = 0.03; OR 6.06; 95% CI 1.11–33.04) and 10
PPI users (45%) vs 5 non-users (17.8%) presented with insulin resistance (p = 0.03, OR 3.8;
95% CI 1.06–13.7) (Supplementary Table 1S).

Discussion

In the present study, we were able to confirm and strengthen our previous findings [6] by
enrolling a larger sample of 301 participants. We found that after 1 year of GFD treatment 72
patients in our study developed MS (4.3 vs 23.9%; p\0.001, OR 6.9) and 112 developed HS
(25.9 vs 37.2%; p\0.01, OR 1.69) (Fig. 1). More significantly, by carrying out a strong
multivariate logistic regression Indication and type of proton pump inhibitor (PPI) prescribed
analysis, we demonstrated that high BMI at baseline (OR 10.8; p\0.001) and PPI exposure
(OR 22.9; p\0.001) were the only independent factors associated with the development of MS
and that HOMAIR (OR 9.7; p\0.001) and PPI exposure (OR 9.2; p\0.001) were the only
independent factors associated with the development of HS. PPI exposure significantly
enhanced MS and HS occurrence in CD patients at 1 year after starting the GFD.
Interestingly, our results appear to be convincing since no difference was seen between PPI
users and non-users at baseline in terms of metabolic features (Tables 1, 5, 6).

Furthermore, we were able to shown that PPI use can induce MS and HS also in a small
sample of patients suffering from GERD, strengthening our results. In effect, PPI exposure
was significantly associated with subsequent MS development, both in CD and in GERD
patients, although the former group had double the risk (OR 6.9 and 3.2, respectively),
probably because of the additive and direct effect of GFD.

With regard to the primary aim of our study to eval uate the predictive factors for MS and HS
in CD after 1 year of GFD it is not surprising that BMI appears to be a risk factor for the
development of MS. BMI can be considered indirectly linked to waist circumference, which
is an intrinsic, fundamental variable in MS.

As mentioned above, the increase in body weight could be due to the improvement in
intestinal absorption (caused by the exclusion of gluten from the diet) in subjects who are in a
compensative hyperphagic state. Several studies have confirmed that long term GFD may not
be nutritionally balanced [6]. Indeed, there is clinical evidence indicating high intake of
simple sugars, proteins and saturated fat and intake of complex carbohydrates and fibre in
such diets [6]. In addition, increased total caloric intake, the macronutrient composition of the
diet, may be involved in the pathogenesis of overweight and obesity in patients with CD.
Many gluten-free foods are characterized by a glycaemic index which is higher than that of
equivalent gluten containing foods [6].

We also found that exposure to PPIs for longer than 6 months after CD diagnosis, in addition
to GFD, was directly associated with the development of MS. Proton pump inhibitors are a
class of drugs commonly recommended in primary and secondary healthcare to inhibit gastric
acid secretion. Although PPI therapy is a successful approach for a variety of conditions,
including GERD, Barrett’s oesophagus and peptic ulcer disease, recent studies have called
into question prescribing practice and safety [26]. As the low pH of the stomach is a vital
defence against pathogens, there is increasing concern that acid suppressing therapy may
have adverse effects. PPI exposure is linked to an increased risk of infectious enteric
conditions, such as a twofold increased risk of Clostridium difficile infection, and to a higher
frequency of community and acquired pneumonias [26–28], nutritional deficiencies and
increased risk of bone fracture [29].

However, data on the effects of PPI exposure on metabolic complications are still scarce. In
the study by Yoshikawa et al. [12], initiation and long-term use of PPIs was associated with
weight gain of at least 5 in 36% of patients using PPIs for GERD and only 4% of patients
who were not using PPIs (over a 2-year period of follow-up). Interestingly, these results are
in accordance with those found in our control group (even if after a retrospective analysis).
A recent US study involving 3073 participants [30] found that PPI use was associated with a
significant weight gain in men and a non-significant weight gain in women, but the authors
did not find evidence of significant differences in energy intake or markers of energy
expenditure. One possible explanation for how PPIs may influence weight is through
increased energy extraction. PPI use has been associated with changes in the microbiota
phenotype; more specifically, with an increase in the relative abundance of Firmicutes and a
decrease in the relative abundance of Bacteroidetes, a profile associated with obesity and
increased energy harvest from the diet [14]. Remarkably, this microbiota phenotype is similar
to that found in patients with CD on a GFD [31].

A recent study by Lin and colleagues [32] demonstrated that Drosophila melanogaster flies
and mice consistently exhibited an obese-like phenotype accompanied by accelerated aging
in flies when gut acid production was suppressed by either genetic or pharmacological means.
These studies suggest that gut microbiota plays a central role in the pathogenesis of
overweight, obesity, metabolic syndrome and hepatic steatosis in patients exposed to PPIs. It
has also been demonstrated that in germ-free mice colonization with the gut microbiota of
obese mice resulted in the ‘‘transmission’’ of obesity [33].

In this setting, it is well known that acid suppressive therapy, in addition to induce a
glycometabolic profile associated with obesity and increased energy harvest from the diet,
may trigger small intestinal bacterial overgrowth (SIBO), a condition that also could play a
relevant role in NAFLD induction/progression [34].

In a study including a wide population of long-term PPI users, Clooney et al. [35] were able
to detect an increased ratio of Firmicutes to Bacteroidetes among PPI users, as well as
significant differences in the abundance of other genera and species between long-term PPI
users and non users.

Imhann et al. [36] found that women used PPIs more often than men did and that PPI users
were generally older (51.6 vs 44.4 years, p\0.01) and had a higher BMI (26.9vs 24.9, p\0.01)
than non-users. These findings indirectly support our results. In line with Imhann et al.’s
theory [36], we hypothesized that long-term exposure to PPIs changes the gut microbiota
through the drugs’ effect on gastric acidity. This hypothesis correlated well with our findings:
we found that HOMA-IR (OR 9.7; p\0.001) and PPI exposure (OR 9.2; p\0.001) were the
only factors associated with the development of HS. In our case, it was not surprising that
HOMA-IR is a predictive factor for HS. Insulin resistance is considered the key factor in the
pathogenesis of MS, of which NAFLD is the hepatic aspect. However, we found a twofold
increase in the occurrence of MS also in patients without PPI exposure (from 3% at CD
diagnosis to 6.4% at follow-up), although this result did not reach statistical significance.
Probably, the use of PPI might enhance the inherent metabolic risks of GFD.

Our study has some limitations that we now briefly discuss. Firstly, ours was not an
intervention study so the ‘‘sample size’’ of the population was not calculated in advance.
However, we are convinced that our sample of approximately 300 patients affected by CD
and prospectively followed up for 1 year offers relevant information about the potential
factors associated with MS and HS in patients with CD on GFD. Secondly, obesity and MS
seem to increase in the general population similarly to how they do in patients with CD [37]
as shown in our control group. However, data in the general population need to be further
elucidated because the sample size of the control group appeared too small. Finally, the
composition of the gut flora of patients with CD was not examined in this study. Further
research is needed to confirm our findings, possibly by evaluating the changes in the gut
microbiota in patients taking PPIs. However, practical limitations for the analysis of
microbiota are well known [30].

order to not affect the observational nature of our study. It would also be interesting to know
if our findings in a population in which the development of MS and HS is faster than in
healthy subjects because of intrinsic features of CD would also apply to the general
population, although several studies and meta-analyses have already demonstrated that PPI
treatment is associated with the risk of cardiovascular events in CD and in the general
population [38–41].

In our study, the diagnosis of HS was primarily based on ultrasound findings, not on
histology (for ethical reasons) or on controlled attenuation parameter (CAP). However, a
recent meta-analysis [42] showed that ultrasonography is an accurate imaging technique in
detecting fatty liver compared to histology, with a sensitivity of 84.8% and a specificity of
93.6% for detecting at least 20% to 30% steatosis. Ultrasonography is relatively inexpensive
and accessible, compared to other diagnostic techniques, and is the imaging technique of
choice for the screening of fatty liver in clinical settings and population studies.

The assessment of compliance to diet and caloric intake was made on the basis of CD
serology, compliance questionnaires [19] and food questionnaires with a patient tailored
approach on the basis of daily caloric needs. Although establishing the compliance to diet and
caloric intake in CD patients is always challenging for physicians, we decided to use these
simple and validated approaches [43].

In conclusion, patients with CD on GFD are at risk of metabolic syndrome and hepatic
steatosis, and this risk is significantly higher with PPI exposure. The use of PPI may enhance
the inherent metabolic risks of GFD. We also suggest that health professionals avoid
prescription of PPIs over prolonged periods of time in patients with CD, reserving this
treatment to the cases in which it is strictly indicated.